Clinical Outcomes of Combination Versus Monotherapy for Gram Negative Non-HACEK Infective Endocarditis

Abstract

The objective of this single-center, retrospective cohort study, was to identify whether combination therapy is associated with a lower rate of adverse outcomes for the treatment of Gram negative non-HACEK IE. The primary endpoint was a composite of 60-day all-cause mortality, readmission, or recurrence of bacteremia. Of the 60 patients included, 56.7% met the primary composite outcome, with 20% overall mortality at 60 days. There was no difference in the primary composite outcome of 60-day readmission, infection recurrence or mortality between groups, with 62% of patients in the monotherapy group and 50% of patients in the combination therapy group experiencing the composite outcome (p=0.36). Despite the high mortality and complicated nature of non-HACEK Gram negative IE, this study showed no difference in 60-day bacteremia recurrence, readmission or mortality among patients treated with combination therapy or monotherapy, suggesting that monotherapy may lead to similar clinical outcomes.

1. Introduction

Gram negative bacilli are uncommon causes of infective endocarditis (IE) due to low affinity for the endocardial tissue compared to Gram positive microorganisms [1]. While HACEK organisms (Haemophilus species, Aggregatibacter species, Cardiobacterium hominis, Eikenella corrodens, and Kingella species) are thought to be implicated in most cases of Gram negative IE, non-HACEK pathogens are found among patients with nosocomial exposures, prosthetic valves, intravascular catheters, people who inject drugs (PWID) or immunocompromised status [1–3,5].

Non-HACEK Gram negative IE is associated with high mortality, with rates greater than 20% in published literature [1, 4–5]. The preferred treatment method for non-HACEK IE is largely understudied and limited to the report of small retrospective trials [2]. In consideration of the limited evidence, weighed against the high mortality rate, the current guidelines from the American Heart Association (AHA) and Infectious Diseases Society of America (IDSA) recommend combination therapy with a beta-lactam and either an aminoglycoside or fluoroquinolone for a six-week course [4].

Given the paucity of literature, we sought to compare clinical outcomes in patients receiving combination therapy (CT) versus those retained on monotherapy (MT) for the treatment of non-HACEK Gram negative IE.

2. Material and Methods

2.1. Study Design

This single-center, retrospective cohort study was conducted at a tertiary care academic medical center in the United States. Patients included were 18 to 89 years old with an inpatient encounter from November 1, 2011 to May 31, 2019 with a positive blood culture for one or more aerobic non-HACEK Gram negative pathogen(s) and an IE diagnosis code. Encounters were identified by query of discharge billing diagnosis codes for IE (International Classification of Diseases [ICD-9 or ICD-10]). The diagnosis was confirmed by physician documentation of IE in the electronic medical record. Cases where IE was suspected without clear documentation were adjudicated by an infectious diseases (ID) physician, based on Duke Criteria for possible or definite IE. Exclusion criteria included prisoners, relapse encounters, and patients with polymicrobial Gram positive or HACEK pathogen(s) isolated from a single blood culture during the same hospitalization.

The primary endpoint of this study was a composite outcome of 60-day all-cause mortality, readmission or recurrence of bacteremia. Secondary endpoints included time to blood culture clearance, presence of embolic events after starting targeted antimicrobial therapy and inpatient mortality. Other variables of interest included adverse events during therapy and multidrug resistant phenotype, based on the Centers for Disease Control (CDC) definition of multi-drug resistant organisms (MDRO) [6]. Selected adverse drug events based on chart documentation during the course of therapy included neutropenia, acute kidney injury, hepatotoxicity, QTc prolongation (>500 ms), glucose disturbances, diarrhea, rash and/or Clostridioides difficile infection. Study data were collected and managed using REDCap (Research Electronic Data Capture) tools at The Ohio State University Wexner Medical Center. REDCap is a secure, web-based software platform designed to support data capture for research studies, providing 1) an intuitive interface for validated data capture; 2) audit trails for tracking data manipulation and export procedures; 3) automated export procedures for seamless data downloads to common statistical packages; and 4) procedures for data integration and interoperability with external sources [7–8].

This study was approved by the institutional Office of Responsible Research Practices Institutional Review Board (IRB).

2.2. Definitions

Combination therapy (CT) was defined as receipt of at least 5 days of two or more antimicrobial agents active against the isolated pathogen. Patients who received less than 5 days of CT were included in the monotherapy (MT) group. Bacteremia recurrence was defined as any positive blood culture for the same initial pathogen following a finalized negative blood culture within 60 days of initiating treatment. Encounters excluded due to relapse infection were defined as a history of IE with the same or other pathogen within the preceding six months.

2.3. Statistical Analysis

Descriptive statistics were calculated for each variable. Quantitative variables were compared using the Wilcoxon rank-sum test. Qualitative variables were grouped into binary or categorical outcomes and compared using the Pearson chi-square test or Fisher’s exact test, as appropriate.

The primary endpoint composite was compared between the MT and CT groups. A multivariable logistic regression analysis was performed to examine the association between the therapy a patient received and their risk of any of the variables included in the primary composite endpoint, while controlling for proven confounders. Variables were considered for inclusion if they were associated with both the outcome and the exposure (p <0.2). A forward selection method was completed within SAS statistical software which included confounders if they changed the intercept between exposure and outcome by greater than 15 percent.

P-values <0.05 were considered statistically significant. Statistical analyses were performed using SAS 9.3 (SAS institute, Cary, NC).

3. Results

A total of 1036 patients with a diagnosis code for IE were screened, with 60 of these patients ultimately meeting all inclusion criteria (Figure 1). The patient population was well-distributed across the CT and MT groups in terms of comorbid conditions, race and ICU admissions (Table 1). Baseline infection and diagnostic characteristics are compared in Table 1. Antimicrobial regimens, durations and adjunctive therapies are summarized in Table 2 and Table 4.

Figure 1: Study Population.

ICD, International Classification of Diseases; IE, infective endocarditis; CLABSI, central line-associated blood stream infection

Table 1.

Baseline Demographics and Disease Characteristics

	Total (n=60)	MT (n=34)	CT (n=26)	P-value (CT vs. MT)
Male sex	40 (67)	19 (56)	21 (81)	0.04
Age	49.5 [35.5–61.5]	59.5 [44–66]	39 [33–56]	0.003
Race				0.85
White	47 (78)	27(79)	20 (77)
African American	10 (17)	5 (15)	5 (19)
Unknown/Not Reported	3 (5)	2 (6)	1 (4)
Intensive Care Unit Admission	27 (45)	13(38)	14 (54)	0.23
Infectious Diseases Consult	58 (97)	32 (94)	26 (100)	0.5
Charlson Comorbidity Index	3 [1.5–5.5]	4 [1.0–8]	3 [2.0–4]	0.27
Comorbidities
Diabetes	15 (25)	9 (26)	6 (23)	0.76
Acute renal failure	16 (27)	10(29)	6 (23)	0.58
Chronic renal failure	17 (28)	11 (32)	6 (23)	0.43
Dialysis	5 (8)	2 (6)	3 (12)	0.64
COPD*	8 (13)	7 (21)	1 (4)	0.12
Cirrhosis	2 (3)	2 (6)	0	0.5
Cancer	5 (8)	5 (15)	0	0.06
PWID*	24 (40)	9 (26)	15 (58)	0.01
Immunocompromised	8 (13)	5 (15)	3 (12)	1
Cardiac Prosthetic Device	25 (42)	13(38)	12 (46)	0.54
CIED*	12 (20)	7 (21)	5 (19)	0.9
Prosthetic valve	15 (25)	7 (21)	8 (31)	0.37
Aortic	10(17)	5 (15)	5 (19)	0.64
Mitral	7 (12)	3 (9)	4 (15)	0.45
Pulmonic	1 (2)	0	1 (4)	0.43
Tricuspid	1 (2)	1 (3)	0	1
VAD*	1 (2)	0	1 (4)	0.43
Echocardiogram				0.3
TTE	27 (45)	17 (50)	10 (38)
TEE	31 (52)	15 (44)	16(62)
None	2 (3)	2 (6)	0
Vegetation Size				0.05
< 10 mm	8 (13)	4 (12)	4 (15)
> 10 mm	23 (38)	9 (26)	14 (54)
Unmeasured	29(48)	21 (62)	8 (31)
Cardiac Involvement				0.03
None Identified	7 (12)	7 (21)	0
Any of the valves	43 (72)	20 (59)	23 (88)
Device Infection	5 (8)	4 (12)	1 (4)
Valve and device	5 (8)	3 (9)	2 (8)
Suspected Source of infection				0.08
Genitourinary	10 (17)	8 (24)	2 (8)
Pulmonary	3 (5)	2 (6)	1 (4)
Gastrointestinal	11(18)	8 (24)	3 (12)
Skin/Soft Tissue	1 (2)	0	1 (4)
Bone/Joint	2 (3)	1 (3)	1 (4)
Surgical	1 (2)	1 (3)	0
Polysubstance abuse	21 (35)	7 (21)	14 (54)
Central line associated	3 (5)	3 (9)	0
Unknown	8 (13)	4 (12)	4 (15)
Pathogen
E. coli	10 (17)	10 (29)	0	0.003
K. pneumoniae	6 (10)	6 (18)	0	0.03
P. aeruginosa	22 (37)	8 (24)	14 (54)	0.02
Serratia spp.	12(20)	3 (9)	9 (35)	0.02
Proteus spp.	2 (3)	1 (3)	1 (4)	1
Stenotrophomonas spp.	1 (2)	1 (3)	0	1
Enterobacter spp.	6 (10)	5 (15)	1 (4)	0.22
K. oxytoca	3 (5)	2 (6)	1 (4)	1
Other^	4 (7)	4 (12)	0	0.13
Secondary Pathogen	10 (17)	6 (18)	4 (15)	1
MDRO Pathogen	16 (27)	12 (35)	4 (15)	0.14
Distant sites of infection
None	35 (58)	26 (76)	9 (35)	0.001
Central Nervous System	10 (17)	1 (3)	9 (35)	0.002
Endophthalmitis	1 (2)	0	1 (4)	0.43
Pulmonary	8 (13)	3 (9)	5 (19)	0.28
Osteomyelitis/discitis	3 (5)	2 (6)	1 (4)	1
Spinal abscess	2 (3)	1 (3)	1 (4)	1
Other+	9 (15)	4 (12)	5 (19)	0.48

Data are presented as number (percent) or median [interquartile range] as appropriate.

^*COPD: Chronic Obstructive Pulmonary Disease, PWID: Persons Who Inject Drugs; CIED: Cardiac Implantable Electronic Device

VAD: Ventricular Assist Device; TTE: Transthoracic echocardiogram; TEE: Transesophageal echocardiogram

^{^}Sphingomonas spp. (2), Pseudomonas fluorescens group (1), Acinetobacter spp. (1)

⁺Splenic infarct (5), Renal infarcts (1), hepatic infarct (1), septic arthritis/tenosynovitis (2)

Table 2.

Treatment and Outcomes

	MT (n=34)	CT (n=26)	P-value
Bacteremia recurrence within 60 days	3 (9)	5 (19)	0.28
Readmission within 60 days	14 (41)	8 (31)	0.36
Mortality within 60 days	7 (21)	5 (19)	0.9
Cardiac Surgery	6 (18)	5 (19)	0.88
Surgery Performed
Single-valve prosthetic (tissue)	1 (3)	3 (12)	0.31
CIED extraction*	5 (15)	3 (12)	1
Treatment Completed	24 (71)	17(65)	0.91
Treatment Incomplete
Death	5 (15)	3 (12)	0.45
Discharge against medical advice	2 (6)	4 (15)
Other^	1 (3)	0
Duration of Treatment (days)	42 [19–44]	47 [42–62]	0.004
Duration of combination therapy (days)	1 [1–3.0]	42 [25–47]	<.0001
Duration of Bacteremia > 2 days	11 (32)	19 (73)	0.002
Embolic Event after starting therapy	8 (24)	10 (38)	0.24
Location of Discharge (regardless of therapy)			<0.0001
Home/Home Health	18 (53)	4 (15)
LTACH	3 (9)	9 (35)
Skilled Nursing Facility	3 (9)	9 (35)
Rehabilitation Facility	1 (3)	0
Against Medical Advice	2 (6)	1 (4)
Other+	6 (18)	1 (4)
Completed therapy inpatient	1 (3)	2 (8)
Adverse Event	0	5 (19)	0.012
Neutropenia	0	0
Acute Kidney Injury	0	5 (19)	0.012
Hepatotoxicity	0	0
QTc prolongation	0	0
Glucose disturbance	0	0
Diarrhea	0	0
Rash	0	0
C. difficile infection	0	1 (4)	0.43
Inpatient Mortality – n(%)	2 (6)	2 (8)	1
Time to culture clearance	2 [1–2.0]	3 [2–5.0]	0.003
Length of Stay	12 [8.0–19]	20.5 [14–32]	0.003

Data are presented as number (percent) or median [interquartile range] as appropriate.

^*CIED: Cardiac Implantable Electronic Device; LTACH: Long-term Acute Care Facility

^{^}Discharge to hospice without antibiotics (1)

⁺Hospice (4), Death (3)

Table 4.

Antimicrobial Regimens

Monotherapy Regimens	N=34
Penicillin	3 (9)
Piperacillin/tazobactam	3 (9)
Cephalosporin	14 (41)
Cefazolin	1 (3)
Cefepime	6 (18)
Ceftriaxone	6 (18)
Cefdinir	1 (3)
Carbapenem	13 (38)
Doripenem	1 (3)
Meropenem	1 (3)
Ertapenem	10 (29)
Imipenem/cilastatin	1 (3)
Fluoroquinolone	4 (12)
Levofloxacin	3 (9)
Moxifloxacin	1 (3)
Combination Therapy Regimens	N=26
Cephalosporin + Aminoglycoside	10 (38)
Cefepime + gentamicin	2
Cefepime + tobramycin	7
Cefepime + amikacin	1
Cephalosporin + Fluoroquinolone	9 (35)
Cefepime + ciprofloxacin	8
Ceftazidime/avibactam + ciprofloxacin	1
Carbapenem + Fluoroquinolone	5 (19)
Ertapenem + ciprofloxacin	4
Meropenem + ciprofloxacin	1
Penicillin + Fluoroquinolone	1 (4)
Piperacillin/tazobactam + ciprofloxacin	1
Monobactam + Aminoglycoside	1 (4)
Aztreonam + gentamicin	1

Data are presented as number (percent) or median [interquartile range] as appropriate.

Of the 60 patients included, 56.7% met the primary composite outcome, with 20% overall mortality at 60 days. There was no difference in the primary composite between groups, with 62% of the MT group and 50% of patients in the CT group experiencing the composite outcome (p=0.36). Similarly, there was no difference between treatment groups for the individual variables in the composite (Table 2). Additionally, no significant difference was observed between groups after controlling for proven confounders in a multivariable logistic regression model (Table 3). Of the 11 patients that received cardiac surgery, 45.5% experienced the primary outcome.

Table 3.

Multivariable logistic regression model for composite 60-day mortality, bacteremia recurrence or readmission

	OR	95% CI
Combination therapy	0.45	0.13–1.6
History of injection drug use	1.78	0.56–5.8
Pseudomonas aeruginosa	1.43	1.15–15.2

CI, confidence interval; OR, odds ratio

The following variables were assessed for inclusion in the model but not retained secondary to p-value >0.2: sex, age, vegetation size, and cardiac involvement. The model was re-ran allowing for an interaction assessment between Pseudomonas aeruginosa and injection drug use. It was determined that the relationship between the interaction term and the outcome was not statistically significant (Adjusted R² 0.159; p-value 0.77).

All patients in both groups achieved clearance of blood cultures; however, the time to culture clearance was longer in the CT group (3 days) compared to the MT group (2 days; p=0.003; Table 2). Notably, there were no significant differences in inpatient mortality or infectious embolic events after starting therapy. Nine patients in the MT group received two antimicrobial agents for less than 5 days of their total treatment course (median 1 day; IQR 1–3 days; Table 2).

The median length of stay was longer in the CT group compared to the MT group (20.5 vs. 12 days; p=0.003; Table 2). Additionally, the CT group was more likely to have a duration of bacteremia greater than 2 days (73% vs. 32%; p=0.002; Table 2).

Of note, no patients in the MT group experienced an adverse event related to therapy, compared to 5 patients in the CT group (p=0.012; Table 2). The primary event was acute kidney injury, in which all patients received an aminoglycoside.

4. Discussion

This study suggests CT is similar to MT, with respect to 60-day bacteremia recurrence, readmission, or mortality in patients with non-HACEK Gram negative IE. This finding persisted after controlling for injection drug use and P. aeruginosa using multivariable logistic regression. Our findings are consistent with those of Morpeth et al. [2], which is the only study to our knowledge to compare outcomes among patients treated with CT or MT.

Overall, 56.7% of patients in our study met the primary composite outcome, with 20% mortality at 60 days. The high mortality rate observed in our study is similar to findings from other observational studies [1,2,9]. This suggests that patients with non-HACEK Gram negative IE are likely to have poor clinical outcomes, regardless of the therapy regimen provided. Cardiac surgery is recommended as an adjunct to antimicrobial therapy in both the AHA/IDSA and European Society of Cardiology guidelines for the management of non-HACEK Gram negative IE, as it may reduce mortality risk [4,10]. In our study, 11 patients received cardiac surgery, 5 (45.5%) of which met the primary outcome, regardless of treatment allocation. A recent retrospective study of 43 patients with non-HACEK Gram negative IE reported similar findings for 10 patients that received cardiac surgery [9]. Six patients (46%) met their primary composite outcome of mortality or infection-related readmission within 90 days. These findings suggest high mortality and readmission despite receipt of cardiac surgery. However, the decision for surgery is complex and often based on several patient factors, which may include patients with complicated, severe disease and high risk of mortality with or without surgical intervention.

In the cohort examined by Veve et al., seventy-six percent of patients received CT, which was most commonly a beta-lactam and aminoglycoside (50%), followed by a beta-lactam and fluoroquinolone (34%) [9]. Notably, three patients (9%) that received an aminoglycoside-based regimen in their study stopped treatment due to development of acute kidney injury. In their multivariate regression analyses, fluoroquinolone-based CT regimens and septic shock were the only factors independently associated with the composite poor outcome. Our study noted higher fluoroquinolone use in CT regimens, whereas previous studies have noted more aminoglycoside-based regimens [2,9]. The fluoroquinolone association with poor outcomes observed by Veve et al. [9] may suggest that fluoroquinolone therapy is inferior to aminoglycoside combinations; however, patients with comorbidities predisposing them to poor outcomes with long-term aminoglycoside therapy may selectively receive fluoroquinolone-based regimens, which could have confounded this finding. Additionally, we also observed acute kidney injury in five patients receiving aminoglycoside-based CT regimens; therefore, the risk of adverse events should be weighed against any potential for poor clinical outcomes.

Patients in the CT group were noted to have longer duration of bacteremia, supported by bacteremia persistent beyond two days. Similarly, we observed a longer duration of therapy and longer hospital length of stay among CT group patients. Notably, the CT group included a larger percentage of patients with CNS involvement and other metastatic sites of infection compared to MT. Additionally, patients in the CT group were more likely to have vegetations greater than 10mm and use intravenous drugs. These findings suggest that the significant differences between groups in terms of bacteremia and treatment duration were related to severity and complexity of infection.

There are several limitations to this study. Based on the nature of a retrospective review, we were unable to account for all factors that may influence the decision to select CT versus MT. Our study showed numerically less patients met the primary outcome among the CT group; however, the difference was ultimately not significant. Due to the rarity of disease, our sample size was small and thus may potentiate the risk of a type 2 error. Additionally, the small sample size limited our ability to identify and control for possible unknown confounders, as well as our ability to perform subgroup analyses, such as for patients that received surgery, where it is difficult to draw clinical conclusions. To balance an adequate sample size with the necessity of clinically useful comparisons, we chose a cut-off of at least five days of CT to be included in the CT group, intending to control for situations where two antimicrobials may have been used empirically while awaiting antimicrobial susceptibility testing. While this cut-off may have potentiated variation in the total duration of therapy within the CT group and limited clinical comparisons, ultimately all patients in the CT group received both agents for at least half of their treatment course. We used all-cause rather than infection-related outcomes, due to the fact that we were reliant on chart documentation for these occurrences and did not want to introduce subjective interpretation. Therefore, it is possible that cases of mortality or readmission in this study could be related to other causes beyond IE. Additionally, risk of mortality associated with infective endocarditis has been previously reported to be increased up to a year post-infection [11–12]. As our study only included patient outcomes up to 60-days, we were unable to assess the effects of CT for a one-year period. Furthermore, this study included patients over the span of several years, in which individual physicians and physician practice may have changed within the institution. Severity of illness at presentation may have affected treatment allocation, as we noted more intravenous drug use, CNS involvement and metastatic sites at baseline in the CT group. However, metastatic sites of infection, area of cardiac involvement, vegetation size and age were not independently associated with the outcome in univariate analyses. Despite this, it is possible that outcome differences between the MT and CT groups were not observed due to a more severe disease state in the CT group. Lastly, the primary identification of subjects was dependent on ICD-9 and ICD-10 coding which may be subject to misclassification; however, we excluded 28 patients (27%) due to improper coding for IE.

5. Conclusions

Despite the high mortality and complicated nature of non-HACEK Gram negative IE, this study showed no difference in 60-day bacteremia recurrence, readmission or mortality among patients treated with CT or MT, suggesting that MT may lead to similar clinical outcomes. A large multicenter cohort study or randomized trial comparing CT with MT would be beneficial to corroborate these findings.

Highlights.

• Non-HACEK Gram negative endocarditis is associated with high mortality

• More than one agent is recommended to treat non-HACEK Gram negative endocarditis

• Combination therapy was not associated with a significant difference in outcomes